Chasing the latest web craze - debate

[Gustavo 'Gus' Carreno]

[Debate Split from https://forum.lazarus.freepascal.org/index.php/topic,44525.0.html]

Hey all,

I quite understand the feeling of all my fellow programmers who want to chase the current status quo of development, the web.
I even understand the ones that fall into the FOMO aspect of things.
I do, don't get me wrong, I do.

What puzzles me is that, even with all this trend to have apps in HTML5+CSS+JS, independent of what generates it, we're seeing a trend to have the newly minted moniker of "Desktop Apps" based on Electron.

Am I that old that, to me at least, the new is just imitating the old, and we're almost full circle into realizing that native compiled "Desktop Apps" should be the way to go?
Is it just me, or is it that because JavaScript devs have been given a hammer they think all solutions are nails?

I dunno, what do you think?

Cheers,
Gus

[lainz]

Hey all,

I quite understand the feeling of all my fellow programmers who want to chase the current status quo of development, the web.
I even understand the ones that fall into the FOMO aspect of things.
I do, don't get me wrong, I do.

What puzzles me is that, even with all this trend to have apps in HTML5+CSS+JS, independent of what generates it, we're seeing a trend to have the newly minted moniker of "Desktop Apps" based on Electron.

Am I that old that, to me at least, the new is just imitating the old, and we're almost full circle into realizing that native compiled "Desktop Apps" should be the way to go?
Is it just me, or is it that because JavaScript devs have been given a hammer they think all solutions are nails?

I dunno, what do you think?

Cheers,
Gus

Is not only JavaScript. Is that you get a better styled application.
Try to make all the components a web can offer, I think you can't and also me that did some bgracontrols for LCL can't.

Animations, full HighDPI Support without problems, scalable graphics, internationalization. All made really simple without the need to compile, just reload the app in the fly. That's faster developed stuff in less time. Every time you hit compile in the Lazarus IDE I already I'm viewing the changes I did in the live reload website.

The same is going for Android with Flutter live reload.

Of course for low level stuff and limited hardware is not good, there is the place for Lazarus. But for the rest I don't see the point of not using HTML+CSS+JS.

[Gustavo 'Gus' Carreno]

Hey lainz,

Is not only JavaScript. Is that you get a better styled application.
Try to make all the components a web can offer, I think you can't and also me that did some bgracontrols for LCL can't.

Let me see if I understand your point: Unless someone can do the presentation layer the way a browser can, we're stuck with something subpar on the logic layer, but hey, we have a magnificent presentation layer, right?

Animations, full HighDPI Support without problems, scalable graphics, internationalization. All made really simple without the need to compile, just reload the app in the fly. That's faster developed stuff in less time. Every time you hit compile in the Lazarus IDE I already I'm viewing the changes I did in the live reload website.

Yeap, you bet it's made simple, well kinda right?
No one is using pure JavaScript, right? How many frameworks are there after jQuery got out?
How flat or steep is the learning curve for those frameworks?
Of all these frameworks and shiny things, how many standards have emerged? None?
Of all those frameworks that have died, how much code can we build upon to make it better without having to re-invent the wheel? None?
But hey, let's sacrifice standards and OS supported API's because... How did you say it? Ah yes "Is that you get a better styled application."

And then you say "That's faster developed stuff in less time. Every time you hit compile in the Lazarus IDE I already I'm viewing the changes I did in the live reload website.".
I admit that the false sense of productivity that an interpreted language gives you feels real. But it's not!!
But explain to me how is that most of the interpreted languages are now trying to ditch the dynamic types in favour of static types, you know, like TypeScript and even the newest of PHP(8 and up)?
Well, it's probably because someone has finally understood that the immediate reward of pressing F5 on a Browser is crap when a client is on the phone every 5 minutes because their site is down because of a stupid mistake due to dynamic types. And then no one can find out to fix it because they don't think about dynamic types. Do you want me to link to the widespread hilarious examples of that in JavaScript?
The fact that when you have static types, you have less error prone code out of the gate.
The fact that the coder that falls for the rapid reward of F5, won't implement tests and while it looks good, it performs like crap.
Should I go on?

The same is going for Android with Flutter live reload.

So the main issue is that we need to evolve the widget set into something more HMTL5+CSS and replace the JS with a compiled language, Dart...
So you agree with me under the condition that we have a dynamic presentation layer that comes from a server somewhere.
So again, you're focusing on the presentation layer. Why is the presentation layer that important?
And what happens when the compiled part of that can't cope with the dynamic presentation layer? Don't we have to swap that?

Of course for low level stuff and limited hardware is not good, there is the place for Lazarus. But for the rest I don't see the point of not using HTML+CSS+JS.

Well, if you give me a solution where someone invents a widget set that can be made by pure HTML5+CSS and then we just have to attach the same events to ObjectPascal instead of JS, will you be happy?

You see, at the moment the HTML5+CSS+JS widget set landscape is a bit of a wild west.
There are no standards. There are no widget sets per se. There's a new framework being invented every day.
I've lived through jQuery. Then it was, what? <insert list of JS frameworks that have been popular since jQuery and ending in Vue>. When is this wild west gonna end?
With all this early adopter and AHHH SHINY attitude it's a no brainer that clients aren't really adopting it. On the contrary, they're getting quite frustrated that past 1 year, their website isn't maintainable because the fad is gone and no one wants to update "yesterday's code".
In comparison, in the present, we have the Windows widget set, GTK, QT and KDE. That's 4 sets, over a period of how many years?

My whole argument is that you shouldn't tear the whole house down just because the presentation layer needs a revamp.
And the Browser is getting to the point where it's being used like the Java or .NET runtime. But without the speed or optimizations of a programming language!!
And if you base all your coding in a binary that you need to carry around but is crap at speed and memory management, is the presentation layer THAT important?

Cheers,
Gus

[Gustavo 'Gus' Carreno]

Hey egsuh,

  1) separate server-side work and client-side work first. Or separate purely logical modules and those which need form or visual control elements. Try to use interfaces to access procedure/functions, even though you can directly access procedures/functions in another module.
  2) Define the communication rules between server and client, considering HTML server. 
  3) Separate two programs - server and client, and try to connect from client to server using THTMLclient. You may use synapse, etc.
  4) Think of moving client-side forms to HTML/CSS/Javascript.

Please correct me if I'm wrong in this attempt of mine to generalize and use some technical terms:

Use an MVC abstraction where you have a server side and a thin client.
The server implements the M(Model) and the C(Controller) while the thin client takes care of the V(View).
The thin client can be a Browser or a purpose built one.
Connect the server and the thin client via the HyperText Transport Protocol(HTTP) using some kind of standard like XML or JSON.

Am I getting this right?

If so then why use Lazarus at all?
As professional programmers, don't we agree we should use the best tool for the job at hand?
If the whole thing is based in MVC, isn't RubyOnRails the best option?
Or if you don't like Ruby, why not CakePHP?
Laravel?
Anything that's been used for web in the past 15 years and with a humongous community and a shit tone of documentation?
Aren't those the arguments we should look at when deciding what tool is better? Stuff with sustainability and maintainability?
Yes we're Lazarus fans, yes Object Pascal is a beautiful language, but can we please loop out of the mentality: Have hammer, is all nails, UGH!! (chest thumping)

"It's the fact we don't have to rewrite so much code." you say.
Erm, don't you? Have you ever counted the man hours to swap from a Windows/GTK/QT widget set to the HTML5+CSS+JS?
Have you factored in the amount of time refactoring all your previous code to now accommodate the MC on the server and the V somewhere else?
It may "feel" easy and quick, but how much mullah is it costing the client?
Has this been fully backed up with tests or is this good cuz you have a feeling?

Just as a disclaimer:

I'm old. I started coding in 1985(was 15) but programming in 1997(was 27 out of Uni) and I've lived through a lot of evolving standards.
It was waterfall, then agile, then TDD. Where are we now?
I've also had to work with a lot of project managers/bosses that could only see productivity in the amount of lines of code, and never the whole picture.
I've also had to try and convince bosses that some standards, new at the moment, would make them save money in the long time: Stuff like source version control and tests.
And through all that had to shut up and submit when they thought they knew better. I'm just glad that I was able to bail out before it all collapsed when reality came a knocking.

So I've heard all the buts and all the ifs and all the arguments that "this is so different", but I'm not convinced and still say that you're looking at this with blinders or just obfuscated by the AHHHH SHINY.
Or worse, the client is insisting in the AHHH SHINY and you're just going along for the ride...

I remember, a long time ago, my dad saying to me: Microsoft is here to stay, why are you so intent in wasting you time with this Linux thing?
And now Windows has Linux inside it and GitHub is owned by Microsoft. The majority is doing web related programming and all the servers are in Linux.
Do we really know how long will the current status quo stand?

I'm old and I've been rooting for native "Desktop Apps" to have a comeback because it just makes sense.
Maybe I'll be vindicated, like the Linux thing above, or maybe not. The future is always a tad foggy.
But if I was a betting man, and with the trends I'm seeing, I think we'll get back to some semblance of native instead of so much interpreted...

Cheers,
Gus

[lainz]

What I can only see is that you're against a technology.
If it's the most widespread and most used don't matter.

I'm in the other hand im not against Lazarus and see a point using it.

Don't mind to respond to every thing you said because I will keep using the technology despite you can say.

Try yourself to see the good points that web has. Everyone knows it and if not can Google them.
Google is the main word here a website if I'm not wrong the number one app used in the world with web technologies.

I left this post since I don't want to talk more.

Edit: didn't want to say but seems that I'm the younger in this post. I've not lived the desktop era too much time and yes the web era. So we're different from the start.

[egsuh]

@ Gustavo 'Gus' Carreno

I'm assuming that the original poster of this thread has a lot of programs written in Pascal (maybe I wrong on this --- if then, why ask here?), and is looking for approaches "To avoid having to rewrite a large part of the application". So I'm suggesting one possible approach based on my experience. Isn't he looking for experiences like mine? Or is he looking for most perfect solution for every occasion? If the latter, I have nothing to say, except that such solution does not exist.   

It's his decision whether to use Lazarus, or re-write in PHP or whatever. I'm not in his position to judge what is the best solution for him. 

[Gustavo 'Gus' Carreno]

Hola Leandro,

What I can only see is that you're against a technology.
If it's the most widespread and most used don't matter.

Hell no, I'm not against the technology.
Most of my career was doing PHP backend development.
And some of my career has been doing native compiled solutions.
But because I've done both I've got a better perspective of both and know not to fall from some of their pitfalls.
It also gives me insight on how to choose the best tool for the job.

I'm in the other hand im not against Lazarus and see a point using it.

Well, and to re-iterate, I'm not against Lazarus, I'm in favour of the right tool for the job.
And Open Pascal, compared to PHP in the context of web development will always loose because PHP has better community and documentation on web projects.
Open Pascal has a better community and documentation on native apps.
I'm not even against using PHP to do a desktop app, if possible, it's just the fact that it wouldn't be the best tool for the job, would it?
Why are you taking this as a personal attack? Did I say that you are wrong?
Nopes, I said that we should base our professional choices in logic, not passion.

Don't mind to respond to every thing you said because I will keep using the technology despite you can say.

I hope you do  and I don't mind at all!!

Try yourself to see the good points that web has. Everyone knows it and if not can Google them.
Google is the main word here a website if I'm not wrong the number one app used in the world with web technologies.

Like I said, most of my career has been in the Web industry. I know it's ins and outs. And a big out is trying to compete in the native "Desktop app" world!

I left this post since I don't want to talk more.

Okydokes, it's your prerogative to disagree with me

Edit: didn't want to say but seems that I'm the younger in this post. I've not lived the desktop era too much time and yes the web era. So we're different from the start.

Well, young just means "have not learned enough yet, should learn more", it's not a bad thing. You still have loads of time ahead of you get old and cranky like me

Saludos,
Gus

[Gustavo 'Gus' Carreno]

Hey egsuh,

@ Gustavo 'Gus' Carreno

Please call me Gus, that's why I have the 'Gus' as a pseudo middle name there

I'm assuming that the original poster of this thread has a lot of programs written in Pascal (maybe I wrong on this --- if then, why ask here?), and is looking for approaches "To avoid having to rewrite a large part of the application". So I'm suggesting one possible approach based on my experience. Isn't he looking for experiences like mine? Or is he looking for most perfect solution for every occasion? If the latter, I have nothing to say, except that such solution does not exist.   

Like you I'm not sure if he's asking here due to having a big code dependency on Object Pascal, or just because he doesn't have the expertize in anything else.
And I should've thank you for your input when I first answered. For that I'm very sorry and will now thank you very, very much for sharing your experience.

That, nonetheless, doesn't invalidate me giving advice on what I consider to be the best tool for the job, based on community size, documentation, sustainability and maintainability of a web project.
And based on that, Object Pascal is not at the top, sorry.

This nonetheless does not invalidate your effort. I'm glad you've been able to dish out your solution in Object Pascal, and would love to read a full tutorial on how you've done it.
If you're ever gonna be interested in doing so!

It's his decision whether to use Lazarus, or re-write in PHP or whatever. I'm not in his position to judge what is the best solution for him.

Well, me neither. But, being the opinionated old cranky git that I am, I'll try and pass my experience unto others.
Isn't it the sole purpose of this forum?
Even If sometimes it means we have to say: I'm sorry, your best bet is with something else.
Unless of course it's an experiment, or just a proof of concept, then yeah, go for it and just have fun.
I'm just thinking that this could be for a paying client and then I'm back to the mantra: Best tool for the job!

I hope I'm getting my message across.
I'm not against any of your endeavours in the past.
If you've done it, I'm rather happy for you, really!
It just a sense of responsibility to your client, not your passion or your limitations.

Cheers,
Gus

[Gustavo 'Gus' Carreno]

Hey all,

Let me be clear on something...

When someone says: I have no financial obligation on this and I want to perform a fun experiment of doing X with tool Y, even if it's "un-wise".
I'll say: Heck dude/dudette go for it. I admire your humph and courage!! Please leave behind a nice tutorial so we can all increase our collective knowledge!

When someone says: A client of mine wants me to do X.
I'll say: I hope you consider the fact that if you get whacked by a bus/get fired/have a family emergency/just life in general, the next guy that comes along will be able to pick up your project. And if your client really has to replace you or add another dev, it's not impossible to find another "you".
And for that to be true, the best thing is to have it based on a big community, large acceptance and non stale documentation.
That means no AHHH SHINY and no Have hammer, is all nail type of approach, right?

That's it, that's the clarity I wanted to pass along...

Cheers,
Gus

[PascalDragon]

You see, at the moment the HTML5+CSS+JS widget set landscape is a bit of a wild west.
There are no standards. There are no widget sets per se. There's a new framework being invented every day.
I've lived through jQuery. Then it was, what? <insert list of JS frameworks that have been popular since jQuery and ending in Vue>. When is this wild west gonna end?
With all this early adopter and AHHH SHINY attitude it's a no brainer that clients aren't really adopting it. On the contrary, they're getting quite frustrated that past 1 year, their website isn't maintainable because the fad is gone and no one wants to update "yesterday's code".
In comparison, in the present, we have the Windows widget set, GTK, QT and KDE. That's 4 sets, over a period of how many years?

You are comparing apples with oranges. On the web there is only one, real widgetset: namely what the DOM provides. Everything else is merely an abstraction of that and the web developers have not yet found the best abstraction for the DOM.

My whole argument is that you shouldn't tear the whole house down just because the presentation layer needs a revamp.
And the Browser is getting to the point where it's being used like the Java or .NET runtime. But without the speed or optimizations of a programming language!!

It seems that you are not aware what kind of optimizations browsers do. After all you're able to run 3D games in the browser using JavaScript and WebGL nowadays with high framerate and polygon counts.

Please correct me if I'm wrong in this attempt of mine to generalize and use some technical terms:

Use an MVC abstraction where you have a server side and a thin client.
The server implements the M(Model) and the C(Controller) while the thin client takes care of the V(View).
The thin client can be a Browser or a purpose built one.
Connect the server and the thin client via the HyperText Transport Protocol(HTTP) using some kind of standard like XML or JSON.

Am I getting this right?

If so then why use Lazarus at all?
As professional programmers, don't we agree we should use the best tool for the job at hand?
If the whole thing is based in MVC, isn't RubyOnRails the best option?
Or if you don't like Ruby, why not CakePHP?
Laravel?

And who says that Lazarus/FPC isn't the best job? There are frameworks like mORMot that can help with the backend. And for the frontend you can use the LCL if you have a custom client or you can use Pas2JS with TMS Web Core or my work-in-progress Web Component Library if you want a JS-based solution.

The main advantage of this is that you don't need learn a new language. You can just use what's familiar. I don't care about Ruby, PHP or whatever because I don't want to deal with them. And I want to deal with as less JS as possible as well, thus I use Pas2JS. And a web client is more portable and simpler for users than having to install some binary client.

[Blade]

...There are frameworks like mORMot that can help with the backend. And for the frontend you can use the LCL if you have a custom client or you can use Pas2JS with TMS Web Core or my work-in-progress Web Component Library if you want a JS-based solution.

The main advantage of this is that you don't need learn a new language. You can just use what's familiar. I don't care about Ruby, PHP or whatever because I don't want to deal with them. And I want to deal with as less JS as possible as well, thus I use Pas2JS.

+1 

[lainz]

Let me see if I understand your point: Unless someone can do the presentation layer the way a browser can, we're stuck with something subpar on the logic layer, but hey, we have a magnificent presentation layer, right?

There is a whole thread in this forum talking about it, look for 'modern ui with lazarus'.

Like I said, most of my career has been in the Web industry. I know it's ins and outs. And a big out is trying to compete in the native "Desktop app" world!

I'm using at work Slack, Paymo, Postman, VS Code, all Electron apps. And are the most used, when I choose to don't use the app I use the web application and works the same.

Okydokes, it's your prerogative to disagree with me

Maybe =)

Really, I see your post as an attack at the first time, but the second one was better (in a possitive way) telling you're a web developer, so I think you understand about the tools.

I'm using both web and Lazarus.

And yes the missing part is the UI in Lazarus, despite I know how to create custom UI, say BGRAControls, where I contribute, there is nothing compared to HTML+CSS+JS out there.

And yes the UI is that the end user sees, they don't care about the low level or high level code, if we use objects or not, if we do in pascal or javascript or any language.

[Gustavo 'Gus' Carreno]

Hey PascalDragon,

I don't care about Ruby, PHP or whatever because I don't want to deal with them.

I don't care about bullies that try to shove their beliefs down someone else's throat.
I don't like that in Jehovah's witnesses, I don't like that in Mormons, I don't like that on (Tele)Evangelists, so I don't really care for that in you.

If you're not humble enough to admit that sometimes you're passion technology is limited and need to pass the bar to some other technology, you're not good to any client.
I could say it's greed, but I'll just say it's an overzealous fragile male ego.
We've had enough of that in our community, so please stop being the dude bro that our community is finally realizing is too toxic to have around.

Cheers,
Gus

[Gustavo 'Gus' Carreno]

Hola Leandro,

Really, I see your post as an attack at the first time, but the second one was better (in a possitive way) telling you're a web developer, so I think you understand about the tools.

I'm glad I was able to tone it down and explain, with less passion, what I was going about
Thanks for understanding.

And yes the UI is that the end user sees, they don't care about the low level or high level code, if we use objects or not, if we do in pascal or javascript or any language.

That is true when in the context of an open source project where there is no money involved and you're trying to attract the attention of all kinds of people.

But in the context of a paying client, you want to provide the most efficient way of providing value to his money.
If you get all tangled up in promoting your passion tools just for the sake of something new or something familiar, you're not giving the client good service. You're just stroking your ego, no?
If you wan to be a good professional, you have to be prepared to say no and direct said client to something that will provide that value on his money.
And mainly remember that the project may outlive you. So be kind to the next guy that comes along to service your mess

Cheers,
Gus

[Gustavo 'Gus' Carreno]

Hey PascalDragon,

While you really don't deserve an answer to this post, I'm throwing caution to the wind and just do it anyway...

You are comparing apples with oranges. On the web there is only one, real widgetset: namely what the DOM provides. Everything else is merely an abstraction of that and the web developers have not yet found the best abstraction for the DOM.

If you want to be that technical, and wanna go bare metal, then in an OS there is the pixel and it's colour, alpha, position in the matrix that is your graphic's memory behind your screen.
Ever since Xerox came up with the first Graphical Interface, the rules of what a widget is were clear: A thing within a visual Rect(angle) that you code via events.
From that we've had many ups and downs but we've stabilized on these: Windows, GTK, Qt and whatever KDE uses, that I can never remember.
In light of that, how are we being presented with a new approach to something already solved, on a yearly or biannual cycle?
Now, do we have the DOM limitations to blame? I sincerely don't know. What I do know is that it's been more than 20 years and where still seeing new JS frameworks popping every year...
20 years is 5 years shy of a generation. Isn't that time enough to get the act together?

It seems that you are not aware what kind of optimizations browsers do. After all you're able to run 3D games in the browser using JavaScript and WebGL nowadays with high framerate and polygon counts.

I'm quite aware of the feats being performed inside a browser. I'm not living under a rock.
What I don't understand is why do we have to add another layer of difficulty.
The OS can do all this with a kernel and some drivers. So why do we need WASM/WebGL+Browser+Drivers+Kernel?
As a show off of what you can do in a constrained environment, I say: Well done!! Write us a blog post so we can all learn about it!!
But don't make us have to use it just because you made an effort on something silly and your ego is gonna bruise if we don't!!

And who says that Lazarus/FPC isn't the best job? There are frameworks like mORMot that can help with the backend. And for the frontend you can use the LCL if you have a custom client or you can use Pas2JS with TMS Web Core or my work-in-progress Web Component Library if you want a JS-based solution.

Like I said, I acknowledge all those efforts with a big BRAVO. Well done, very nice to have in my passion language.
But that doesn't blind me to the fact that in comparison to other web technologies, they have a bigger community, better documentation and more people in the recruitment pool.
And in the business world, those are the params that count, not that you love Lazarus and have a solution in it.
Repeating myself: If you get smacked by a bus, will the client find an easy and affordable replacement for your niche set of tools?

Cheers,
Gus